Skip to content

Pathology Tumor Detection#296

Merged
Nic-Ma merged 8 commits intoProject-MONAI:masterfrom
bhashemian:camelyon_ignite
Aug 2, 2021
Merged

Pathology Tumor Detection#296
Nic-Ma merged 8 commits intoProject-MONAI:masterfrom
bhashemian:camelyon_ignite

Conversation

@bhashemian
Copy link
Copy Markdown
Member

Description

This PR adds a pathology pipeline example to train and evaluate a model to detect tumors on whole-slide histopathology images. This pipeline uses ignite-based MONAI workflow.

Status

Ready

Checks

@bhashemian bhashemian changed the title Pathology Pathology Tumor Detection Aug 1, 2021
@bhashemian
Copy link
Copy Markdown
Member Author

@wyli, @Nic-Ma, it would be great if either of you can review and merge this PR. Thanks.

@Nic-Ma
Copy link
Copy Markdown
Contributor

Nic-Ma commented Aug 1, 2021

Hi @drbeh ,

Thanks for the quick PR!
I think the dataset_subset.json is too long in the code base, could you please help put it in our google drive storage? And use MONAI download API to download it in this example?
@wyli , what do you think?

Thanks.

@wyli
Copy link
Copy Markdown
Contributor

wyli commented Aug 1, 2021

@bhashemian
Copy link
Copy Markdown
Member Author

Sure, let me check that.

Signed-off-by: Behrooz <[email protected]>
@bhashemian
Copy link
Copy Markdown
Member Author

bhashemian commented Aug 1, 2021

@Nic-Ma, @wyli, I moved the datasets to the google drive and updated the README with the download links. For testing this pipeline (without any performance gain), you should download two images (tumor_001 and tumor_101) as mentioned in the README from here
You also need the dataset_0_subset_0.json from here.

@wyli
Copy link
Copy Markdown
Contributor

wyli commented Aug 1, 2021

Looks good to me. My concern is that we'll probably break this tutorial when developing the core codebase. This happened before Project-MONAI/MONAI#2571, and it was detected manually -- which is not efficient.

So I'm wondering whether we should also convert this into an integration test in MONAI (e.g. https://github.com/Project-MONAI/MONAI/blob/dev/tests/test_integration_determinism.py), to download some testing data (e.g. https://github.com/Project-MONAI/MONAI/blob/7f42efe0ab55730c302aba87bcd9bf75c4bfc496/tests/test_patch_wsi_dataset.py#L27-L28), run a few training steps.

@bhashemian
Copy link
Copy Markdown
Member Author

@wyli, I agree with you that we should have an integration test in MONAI for pathology too. I can make a simplified version of this for integration but it requires to download two large images (3.1GB and 1.4GB). Wouldn't it be a problem?

Anyways, We should have some examples in tutorials, shouldn't we?

@wyli
Copy link
Copy Markdown
Contributor

wyli commented Aug 1, 2021

Adding tests will be great. gigabytes of testing data will cause issues, is it possible to have a smaller test?

Yes we should have examples in the tutorial, and ideally other developers should be able to help maintain it when you are not available.

Copy link
Copy Markdown
Contributor

@wyli wyli left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

thanks, created a separate ticket for the integration test Project-MONAI/MONAI#2676. @Nic-Ma could you please help confirm/merge this PR?

Copy link
Copy Markdown
Contributor

@Nic-Ma Nic-Ma left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks good to me.
Could you please help also update the README doc of this repo to include this pipeline?
https://github.com/Project-MONAI/tutorials/blob/master/README.md

Thanks in advance.

Signed-off-by: Behrooz <[email protected]>
@bhashemian
Copy link
Copy Markdown
Member Author

README is updated!

@Nic-Ma Nic-Ma merged commit b3130b6 into Project-MONAI:master Aug 2, 2021
@bhashemian bhashemian deleted the camelyon_ignite branch August 2, 2021 15:39
boneseva pushed a commit to boneseva/MONAI-tutorials that referenced this pull request Apr 21, 2024
* Add README

Signed-off-by: Behrooz <[email protected]>

* Add camelyon train/evaluate pipline with ignite

Signed-off-by: Behrooz <[email protected]>

* Update default dataset

Signed-off-by: Behrooz <[email protected]>

* Update README with dataset downloads

Signed-off-by: Behrooz <[email protected]>

* Change the reference to the default dataset

Signed-off-by: Behrooz <[email protected]>

* Update READMEs

Signed-off-by: Behrooz <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants